Personalization and the Web from a Museum Perspective

Jonathan P. Bowen, London South Bank University, United Kingdom; and Silvia Filippini-Fantoni, Université Paris I - Sorbonne, France

http://my.museophile.net

Abstract

Web sites are increasingly adapted towards their users by a variety of dynamic techniques, providing improved personalization for the individual. In this paper we discuss various possible approaches from a museological as well as a technical point of view. A number of Web sites, some produced by museums and some in other fields such as e-commerce or Internet based information services, will be considered as case studies. All use personalization techniques in ways that could be beneficial also for the cultural sector. Other possible applications such as audio guides or interactive devices will also be considered, especially with respect to how this technique could help in the personalization of museum Web sites. Currently most museums lag well behind the state of the art with respect to on-line personalization. Perhaps this paper will inspire some museums to add personalization facilities to their Web sites the next time they undertake major updates.

Keywords: Personalization, customization, adaptivity, adaptability, museums, Web sites.

I want freedom for the full expression on my personality.

- Mahatma Gandhi (1869-1948)

Introduction

Negroponte (1995), in the chapter on his thoughts about The Post-Information Age in his widely acclaimed book Being Digital, noted that:

True personalization is now upon us. It's not just a matter of selecting relish over mustard once. The post-information age is about acquaintance over time: machines' understanding individuals with the same degree of subtlety (or more than) we can expect from other human beings, including idiosyncrasies (like always wearing a blue-striped shirt) and totally random events, good and bad, in the unfolding narrative of our lives.

Of course we are not there yet, but we have made some progress in the best part of a decade since these words were written. Certainly many advanced commercial Web sites have used personalization techniques in increasingly sophisticated ways. In the past few years, personalization through information technology (specifically via the World Wide Web) has become an increasingly significant trend in the museum world too (Filippini-Fantoni, 2003a; Filippini-Fantoni & Bowen, 2004), where it has been introduced to make facilities more relevant and useful for individual users. By providing differentiated access to information and services according to the user's profile, personalization helps to respond to museums' educational, marketing, as well as usability needs.

In this paper, we first explore the origins of personalization in the context of on-line museums. We also consider a wider perspective of personalization in other areas such as e-commerce, where the associated techniques are already used in a relatively advanced manner and where lessons could be learned by museums. However, the issue of why museums should bother with the approach at all is also important. Despite the fact that most museum Web sites do not use personalization techniques, there are some interesting projects already available on-line and others, even more advanced, underway. We survey some of these and the different ways that museums are using or planning to use personalization. Finally we present some conclusions on the progress so far and possibilities for the future. Overall this is an interest area that deserves more attention by the museum community, especially larger museums with significant numbers of on-line visitors.

Note that the term "personalization" and the related term "customization" are sometimes used rather loosely. In this paper we will refer to customization as the ability of the user to modify the interface to individual requirements, whereas personalization occurs when the system rather more surreptitiously adapts the interface for the user, either "explicitly" from personal information provided by the user, or "implicitly" by monitoring the user's actions. We shall expand on this later in the paper. For want of a better collective term, we dub all forms of customization and personalization as 'individualization,' a word that can found in the index of Neproponte (1995). Note that authors of some other papers have used the term 'personalization' in a wider context to include at least some forms of customization as well.

Personalization: Its origins and evolution

Personalization on the Internet has come about as a result of a long evolutionary process accelerated by the rapid development of the World Wide Web in the 1990s (Berners-Lee, 1999). With the Internet and the Web, a new communication tool has been made available, enabling people with different characteristics and goals to access an ever-growing quantity of information for personal use.

However, under these conditions it becomes very difficult for people to find the right information at the right time and at the right level of detail. In order to find a solution to this problem, researchers from different communities have developed systems with an ability to adapt their behavior to the goals, tasks, interests, and other features of individual users and groups of users. The result is what we normally call adaptive or personalized systems/hypermedia. What distinguishes these systems from the traditional "static" Web is the creation of a user model that represents the characteristics of the user, utilizing them in the creation of content and presentations adapted to the different individuals (Brusilovsky & Maybury, 2002). By so doing, personalization becomes a useful tool in the selection and filtering of information for the user, facilitating navigation and increasing the speed of access as well as the likelihood that the user's search is successful.

The principal application area of adaptive Web systems is e-commerce, where personalization has come to symbolize a new production paradigm in contrast with the 20th Century model of mass production. Peppers and Rogers (1996), a marketing book (Pine II, 1999), illustrates the possibility of differentiating the production and sale of products and services according to the customer's profile. Web personalization allows companies to offer products, services and advertising that give consideration to the interests, characteristics and needs of the customer, consistent with the information obtained during the navigation of the user on the company's Web site. E.g., see the Personalization Consortium (http://www.personalization.org), which has an interest in one-to-one marketing technology. Probably the leading example is Amazon (http://www.amazon.com), the on-line bookstore and retailer, which monitors the user's interests through any previous navigation around the Web site.

The  desirability of taking into consideration the specific interests or characteristics of the final users when providing a product or a service is not limited to the business world. Web personalization has also become very popular in sectors other than e-commerce, such as information retrieval (where personalization techniques are used to provide responses more relevant to the user's search queries) as well as medicine, tourism and education.

Personalization has been an increasingly popular approach on the Web in recent years ,and much research and development effort has been expended. Witness, for example, two recent special issues of the monthly Communications of the ACM, a leading computer science professional journal, on Personalized Views of Personalization and The Adaptive Web (Riechen, 2000; Brusilovsky & Maybury, 2002). See, for example, the GUIDE Project (http://www.guide.lancs.ac.uk) that is considering the personalization of a mobile computer-based and context-aware tourist guide using some of the lessons learned from the Web (Chevest et al., 2002).

Among the last to recognize the importance of personalization were cultural institutions such as museums, libraries and archives, which only relatively recently realized the importance of the Internet as a new means of communication with their public. Given the growing number of visitors looking for on-line information concerning their collections and activities, it has become fundamental for these institutions to try to improve their visitors' ability to navigate on-line and to access information in the most effective way. Personalization is a viable and worthwhile aid in this process.

Within the cultural sector, the first personalized Web applications were developed in libraries (Russell, 2003), where, as Lynch (2000) explains, personalization can assist librarians in their role as intermediaries between information and the public. Personalization can in fact be used to assist in the selection and filtering of material for the users, who otherwise would spend too much time in looking for what is needed. Personalized service systems can create models of users' interests and use this to prioritize information and sort search results so that users are able to find interesting material quickly and easily in the library's catalogue. Examples of these types of applications include the My Schoolnet and My Library projects described in Bonnet (2002).

Web Personalization in Museums: Why?

The important role that personalization plays in museums was anticipated by Rosalind Picard, Director of the Affective Computing Research Department at MIT (Picard, 1997), who declared:

Museums and galleries that are becoming available on the World Wide Web could personalize tours and suggest to you additional collections to browse, after observing your reaction to what you have already seen. The idea is to move computers toward more personal service, tailored to your ever-changing interests, without increasing the demands on you to explicitly state your preferences.

Personalized access to collections, alerts, agendas, tour proposals and tour guides are just a few examples of the different applications that have recently been developed by museums all over the world. The reasons for such an incredible affirmation are numerous. Personalization not only has the advantage of improving the usability of a Web site by facilitating navigation and helping people find the right information, by taking the differences in age, level of education, learning style and previous knowledge into consideration when providing information, adaptive systems also have the potential to facilitate the learning process. Visitor studies seem to confirm that learning is stimulated when the information is described in terms that the visitor can understand and if it makes reference to their interests as well as concepts that the visitor has already encountered during the navigation/visit.

Personalized systems help to recreate the human element that "listens to the visitor" and "understands him" by offering a personal touch; this is another important factor that contributes to the success of Web personalization in museums. It is a particularly important element, especially for audio guides, which must offer a certain level of flexibility in order to adapt the contents to the needs and interests of the users, just as a real museum guide would do. It could eventually also help on-line, making the visitors feel comfortable and oriented in the virtual space through the use of adapted virtual avatars.

By facilitating access to the information and the learning process and by recreating the human element normally lost with the use of technology, personalized systems have the potential to satisfy the visitor, who is then more stimulated to come back and reuse the system or to encourage other people to try it as well. This can be economically advantageous for museums, which could expect an increase in virtual and real visitors as a result of personalization.

Personalization: A New Communication Strategy for Museums

Personalization in museums could begin an evolution of the relationship between the visitor and the institution. Museums' traditional communication paradigms often share the characteristic of mass communication experiences, where one of the parties is present and the other absent. With personalization, these can be modified into more natural forms of communications based on interaction and exchange. By considering the different necessities of each individual, it is now possible to move from talking to the visitor to talking with the visitors. In this way the museum monologue turns into a dialogue: the visitors teach the museum more and more about themselves, their interests and needs and, in return the museum proposes content and services that fit. Personalization is therefore a new communication strategy based on a continuous process of collaboration, learning and adaptation between the museum and its visitors.

Types of Adaptation

So far, we have explored how Web personalization can be used as a new communication strategy for museums, where it has appeared in response to educational, marketing and usability needs. Beyond these common goals, however, there is a great diversity in how personalization through IT can be achieved. The techniques available to collect information about users and the methods used to process such information to create user profiles and provide adapted content, presentation and/or structure, are varied. A brief description of the different techniques and methods will be provided here before we move on to illustrate different application cases inside and outside the museums world.

A first important distinction concerning the amount of control the user has on the personalization process can be made between customization and personalization .

Customization or adaptability occurs when "the user can configure an interface and create a profile manually, adding and removing elements in the profile" (Bonnet, 2002). The control of the look and/or content of the site is explicit and user-driven; i.e. the user is involved actively in the process and has direct control. A good example of this type of adaptation is provided in Kobsa et al. (2001):

a user of a Web site may see a need to introduce a shortcut for a webpage that is frequently visited but deeply buried in the site hierarchy. The user then introduces a new link on the lateral navigation bar (if possible) or defines a bookmark in the browser, selects the shortcut link and performs the necessary step to produce this adaptation.

Well-known examples of sophisticated customizable general portal Web sites are available from Yahoo - My Yahoo! - http://my.yahoo.com) and Microsoft MSN (http://my.msn.com). The users are able to register with unique password-protected usernames and then configure the Web site interface and contents to their own requirements by selecting, positioning and customizing a number of boxes on the Web page. Such mainstream Web sites are now well established. For example, My Yahoo! was introduced in July 1996 (Manber et al., 2000), almost eight years ago.

A museum-oriented example is the Museophile accessibility Web service (http://access.museophile.net/index.museum). Here the Musedoma index Web site of museums registered under the .museum top-level domain is presented using a text-only interface implemented using the freely available open source Betsie tool, originally developed by the BBC (http://betsie.sourceforge.net). This is aimed at aiding disabled access on-line (Bowen, 2004). In particular, partially sighted people can select the text foreground/background color, size and font that is most readable for them by following a link at the bottom of every page at any time. The service gives access to all museum Web sites under the .museum domain in a customizable text-only form.

Another museum-related example is the Museophile forums Web site (http://forums.museophile.net). Here a registered user can select from a range of museum-related newsfeeds and mailing lists that are presented on a personalized homepage (Bowen et al., 2003). When the user revisits the site, information stored as a "cookie" (see later) allows the latest items in the selected sources to be displayed on the right hand side of the page.

In personalization or adaptivity, on the other hand, the user is seen as being passive, or at least somewhat less in control (Bonnet, 2002). Modifications concerning the content or even the structure of a Web site are performed automatically by the system based on information concerning the user stored in the so-called user profile. Such information about the user is provided either explicitly, by that user , in on-line registration forms, questionnaires and reviewing (static profiles) or implicitly by recording the navigational behavior and/or preferences of each user through cookies and Web server log files (dynamic profiles) (Eirinaki & Vazirgiannis, 2003). A "cookie" is a small piece of data sent by a Web site and stored on the client-side (browser) computer and can be reused later on the server-side (the Web site that sent the cookie) as unique information concerning a user. A Web server log is record of each access to a Web server with information such as the name of the client computer, the date/time and the resource accessed.

As a simple example of explicit personalization, see the UpMyStreet Web site (http://www.upmystreet.com). Here the user can type in a United Kingdom postcode (e.g., for their home) and obtain personalized local information including a map, property prices, businesses, schools, the weather, local government contact information, statistics, etc. A personalized hyperlink can even be used; for example, see the information for the workplace of one of the authors (http://www.upmystreet.com/overview/?l1=SE1+0AA). We will present more museum-oriented examples of personalization later in the paper.

Once the data concerning the users are collected, implicitly or explicitly or even in both ways, appropriate content is determined and delivered. This process usually follows one or a combination of the following techniques: content-based filtering, collaborative filtering, rule-based filtering and Web usage mining. Content-based systems tracks user behavior and preferences and recommend items that are similar to items chosen in the past. For example, if a user shows an interest in paintings of a particular style or period, or by a particular artist, links to other related pictures can be presented.

Collaborative filtering compares a user's tastes with those of other users in order to build up a picture of like-minded people. The choice of content is then based on the assumption that this particular user will value what like-minded people also enjoyed. The user's tastes are either inferred from previous actions or else measured directly by asking the user to rate products. A typical example of the use of this technique is Amazon (http://www.amazon.com), which determines a user's interests from previous purchases as well as from ratings given to titles (Linden et al., 2003). The user's interests are compared with those of other customers to generate titles that are then recommended during interaction. This is a method echoed by a number of on-line retailers, and it is also used to power recommendation engines for entertainment and television viewing.

A common technique is rule-based filtering, which allows Web site administrators to specify rules based on static or dynamic profiles that are then used to affect the content served to a particular user (Mobascher et al., 2000). For example, association rules could explicitly encode the fact that users who visit two pages may also be likely to be interested in the third related page. More concretely, an interest in paintings by Monet and Renoir could potentially demonstrate a general interest in Impressionism and other Impressionist artists.

Last but not least, there is Web usage mining which relies on the application of statistical and data-mining methods to the Web server log data, resulting in a set of useful patterns that indicate users' navigational behaviors. The simplest methods of analysis that can be applied to such data is statistical analysis which takes input raw Web data and processes this to extract statistical information such as site activity, diagnostic, server, referrers, demographic statistics and click stream analysis. Such analysis has been undertaken on the Virtual Library museums pages (http://icom.museum/vlmp) over several years (Bowen, 2002).

More advanced data mining methods and algorithms for use in the Web domain include association rules (a technique for finding frequent patterns, associations and correlations among sets of items), sequential pattern discovery (an extension of associated rule mining in that it reveals patterns of co-occurrence incorporating the notion of time sequence), clustering (used to group together pages or users that have similar characteristics) and classification (a process that maps items into various classes such as different types of user profile). The patterns discovered through all these techniques are then used to provide personalized content and presentation to users based on their current navigational activity. The personalized content can take the form of recommended links or products, targeted advertisements, or adapted descriptions of objects.

Figure 1. An overview of personalization and customization

Initial Experiments with Web Personalization in Museums: Adaptivity

The first examples of Web personalization in a museum context were developed in the second half of the 1990s in strict relation with the affirmation of academic research on adaptive hypermedia. The ILEX (Intelligent Labelling Explorer) project (http://www.hcrc.ed.ac.uk/Site/ILEXINTE.html), for example, was developed between 1996 and 1997 by the University of Edinburgh in cooperation with the National Museum of Scotland, with the aim of providing intelligent labels for the description of the jewel collection available on the museum's Web site (Hitzeman et al., 1997; Oberlander et al., 1997). By using natural language generation techniques, ILEX, of which only a prototype was made, provided museum artifact descriptions that took into consideration both the level of knowledge of the user and the history of the interaction. Web pages for each object in the museum's collection were generated as the visitor navigated around the system's Web site. These were virtual descriptions in that they did not exist in any form prior to the user's request, but were created on the fly by the system, and adapted to the individual user's needs and behaviors. The result is called dynamic hypertext. Given the wide range of different goals that visitors have, the benefits of dynamic hypertext are important for museums. The production of on-line texts is highly flexible and can be tailored to a level of detail beyond that achievable with static hypermedia. It is possible to introduce a significant amount of variation, presenting each user with object descriptions that are built on a personal level of knowledge, interests and history of interaction.

Adaptivity and dynamic hypertext are also the basis of another early on-line personalized application, designed in 1998 by the National Research Centre: the Carrara Marble Museum (http://giove.cnuce.cnr.it/Museoeng.html) Web site's virtual guide, which accompanies visitors during the exploration of the on-line collection (Paternò & Mancini, 1999). When the agent associated with the virtual guide is activated, the presentation of a work of art in the museum's collection is accompanied by an extra window dedicated to the comments of the virtual guide, which offers additional information to makethe user's visit more interesting and pleasant. This extra information is personalized in the sense that it automatically takes into consideration the works of art, the artists, the materials and the techniques previously seen by the user. The adaptive information that is dynamically generated refers to five different categories of content: introductory, summative, comparative, dissimilar and peculiar. The priorities among information types are managed by a set of rules defined by the programmers (rule-based filtering) that take into consideration the profile chosen by the user when accessing the collection (tourist, student or expert) and the items accessed. Figure 2 shows a particular Web page for a sculpture in the museum.

Figure 2. Carrara Marble Museum Web site screenshot

Customization and Explicit Personalization: A Simpler Approach to Web Personalization

By providing such coherent and contextualized information, modeled on user interaction with the exhibition space as well as with the system itself, adaptive systems have enormous potential in the museum context. However, because of the complexity attached to developing such systems (Filippini-Fantoni, 2003a) and the relatively long interaction that is needed before the system is able to adapt the information efficiently to the user's behavior, museums have (quite sensibly) preferred to introduce simpler forms of personalization for their on-line applications, such as customization or explicit personalization. The above-mentioned kind of individualizations give more control to the user, who either decides the terms of the adaptation and when to employ it (customization) or provides directly the personal information (mainly through fill-in forms) necessary to obtain the adaptation (explicit personalization). Beside the obvious limitations implicit in these techniques, which do not take into account that the visitor's interests and needs might change during the exploration and might demand a reconfiguration of the system, they frequently seem to be employed in museums mainly because they are uncomplicated. The Carrara Marble Museum Web site, for example, besides the adaptive virtual guide, offers an interesting example of customization. Access to the collection can be obtained either by choosing among one of three predefined profiles (tourist, art student and expert), or by customizing a personal profile by defining manually which type of access to the collection (by map, detailed or simple research engine), which type of information, and which type of layout is preferred. Once the profile is manually defined by the user, access to the collection during the on-line session is provided according to the parameters set (Paternò & Mancini, 1999).

This particular system does not provide the visitor with the possibility of saving or recording for future consultations the parameters set. However, this can be easily made possible by allowing the visitor to register with a Web site and to log in again at each visit using a username and password. This can even be automated using the technology of cookies that can record login information on the visitor's local computer, provided this uses a personal and secure file store.

A typical example of these techniques is provided by the Calendar application on New York's Metropolitan Museum of Art Web site (http://www.metmuseum.org/calendar). The principle is very simple: for all those who are bewildered by the amount of daily programming at the Met, the system offers the possibility to set a personal profile indicating particular interests by selecting from the many event types and subjects on a specific form. Every time that visitor returns to the Web to check the events on the daily calendar and clicks into My Met Calendar mode, through the use of cookies only those events that fit the established criteria will be displayed for that particular date or range of dates.

By using the same personalization technique through on-line form registration, the Louvre museum has been working on developing a customized alert system for its new Web site, which will probably be on-line by the end of 2004. Instead of providing everybody with standard information concerning the museum's activities, services and products, this personalized alert system will allow the virtual visitor to set an individual profile by defining personal interests on a fill-in form and providing an email address. Whenever an exhibition, a conference or a concert of interest is about to take place, the user will be automatically informed about it by email or SMS text message.

Personalized visitor plans are also inspired by the same principle. Upon filling in a profile, where the future visitor has to indicate how and when they are tentatively planning to come, with whom and how long they plan to stay, what sort of interest they have or which language they understand, the system will provide a personalized plan for the visit, taking into consideration the submitted parameters. If the visitor is planning to come with children, for example, the system can advise them to opt for specific dates when suitable family activities are scheduled. Specific paths suitable for children will be suggested, together with particular learning material from the on-line bookstore or from the Web site.

Alternatively, if the person is planning to come alone and indicates a specific interest for, let us say, French Impressionism, the system can suggest specific activities (conferences or seminars) or exhibitions on the subject as well as paths that correspond to the interest and time requirements of the visitor. If the user wishes it, the plan can be emailed directly. Personalized museum plans can be very useful, especially for big museums where visitors are likely to be overwhelmed by the number of objects. In such a context, visitors often find it difficult to decide what they want to see or do. Answering a few very simple questions or defining a few criteria on the Web site, and eventually even on site through an interactive device, can help them to overcome these limitations and to enjoy the visit more fully.

A number of museums, including the Louvre and La Cité des Sciences et de L'Industrie in Paris, are working on developing Web-based and on-site applications based on these principles. In particular, the National Museum of Ethnology in Leiden, The Netherlands (http://www.rmv.nl) has developed an on-site application called The tour of the world in 80 questions that allows children aged 7 to 13 to print out a personalized tour plan of the museum based on a personal choice of subjects and continents. The tour plan, which is colorful and easy for children to understand, includes a series of maps that help locate the objects, a brief description of the artifacts, and a list of related questions which the young visitors need to answer during their museum exploration.

Personalized Web Galleries

Another approach to personalization used by museums is the possibility to select images from the digitized collection and to create a personalized Web gallery, accompanied by personal comments or descriptions. The concept was first developed in the early 1990s by Beardon & Worden (1995). They developed a prototype called the virtual curator with the aim of drastically changing the underlying metaphor of the pre-classified exhibition. This allowed visitors to construct an exhibition or display on their own, choosing the objects in which they are most interested and eventually writing comments about them. This project, which initially started as a tool that would enable Design History students to gain the intellectual benefits from curating a small exhibition without all of the resources required to stage a real one, paved the way for a series of initiatives that are now available on a number of museum Web sites.

One of the most worthwhile examples is the Make your own gallery kiosk that premiered at the SFMOMA San Francisco Museum of Modern Art's Points of Departure exhibition (http://www.sfmoma.org/exhibitions/exhib_detail.asp?id=58) in Spring 2001. This allowed anyone to become a virtual curator by creating a virtual exhibition and making it available on-line for other visitors, friends and relatives. The exhibit features works from the museum's permanent collection, arranged to correspond to themes developed by the curatorial theme. By using a mouse to drag images of the exhibition's art pieces into a virtual on-screen gallery and a keyboard to add their own commentary, visitors can essentially re-curate the show and arrange the art according to their own interests.

Other interesting examples that have gained the attention of the academic world are Art Tales, an interactive story-telling program developed by the National Museum of Wildlife Art (http://www.wildlifeart.org/ArtTales), and My Art Gallery, developed by the Seattle Art Museum (http://www.seattleartmuseum.org/myartgallery).These allow visitors to interact with the museums' permanent collections and create their own on-line exhibitions. The Virtual Museum of Canada also allows on-line visitors to interactively create their own personal museum Web site of selected images (http://www.virtualmuseum.ca/English/Personal).

From Personalized Web Galleries to Personalized Pages

The tendency today is to make personalized applications like those above even more complex by adding additional tools that contribute to the creation of an actual personal space within the museum's Web site. This can eventually also be customized according to need: visitors can select links to their favorite sections of the Web site, access their personalized agenda and store links to images, information or articles from the Web site for future consultation or research. This type of application is mainly conceived for frequent visitors or for special categories of visitors who use the Web site as a working tool - such as teachers, journalists, experts, students or researchers. Once the page has been created and eventually customized, visitors can log in every time they access the Web site to find all the information they need, some of which, like the personalized agenda, will be automatically updated by the system every time the dates of relevant new events are established.

The Metropolitan Museum of Art, for example, now provides a complete My Met Museum area on their Web site (http://www.metmuseum.org/mymetmuseum).It includes a My Met Gallery facility to gather favorite works from the museum on-line, Set My Met Calendar, a free newsletter, and a Rapid Check Out option, to facilitate recognition when the person returns to make a purchase in The Met Store (it recognizes personal addresses). Figure 3 shows some screenshots of on-line My Met features provided by the Metropolitan Museum.

Figure 3. "My Met Museum," "My Met Calendar" and "My Met Gallery" facilities

Another example of this type of application is provided by the Ingenious project (http://www.nmsi.ac.uk/nmsipages/nofdigitise.asp) being undertaken in the UK National Museum of Science and Industry museum group. Funded by the UK New Opportunities Fund (NOF), this project, which will be on-line from early Spring 2004, aims at creating a learning environment for the public from the digitized collections of the Science Museum (London), the National Railway Museum (York) and the National Museum of Photography, Film and Television (Bradford) in the UK. Users of the Ingenious Web site will be able to explore and discover the rich collections of these museums through 50 narrative topics, over 30,000 images, and other content-rich resources such as library and object records. In addition, users will be provided with tools for entering a topical debate and personalizing their experience in the so-called CREATE area, where registered users can save images and/or links from the debate areas, read sections and search queries. The users will also be able to send e-cards and create a personal Web gallery from their bookmarked images, and incorporate personal comments that can be emailed to friends and colleagues.

Figure 4 shows a general shot of the Ingenious home page. The facilities include My E-cards to send electronic cards (Figure 5), selected hyperlinks (Figure 6), saved images (Figure 7) and Web galleries (Figure 8).

Figure 4. Ingenious project main page

Figure 5. Ingenious "My E-cards" screenshots

Figure 6. Ingenious "My Links" screenshots

Figure 7. Ingenious "My Saved Images" screenshots

Figure 8. Ingenious "My Web Galleries" screenshots

In other cases, these personal spaces can also include information about a visitor's actual visit in the museum, thus creating a direct link between the visit and the post visit experience. Personalization is an effective tool for stimulating visitors to follow up from home, through a museum's Web site, what caught their attention during the exhibition. For example, the Visite Plus service offered by the Cité des Sciences et de l'Industrie (http://www.cite-sciences.fr/francais/ala_cite/cite_pra/visite+/global_fs.htm) in Paris, which has so far been tested on two of their temporary exhibitions Le Cerveau Intime and Le Canada Vraiment, allows the visitor to configure a personal profile (language, disabilities, etc.) on an interactive kiosk placed at the beginning of the exhibition through a special bar-coded ticket for the former and on a PDA (Personal Digital Assistant) for the latter. This can then be used to access adapted information from the different interactive devices and to play various games and quizzes in the exhibitions. The results of such interaction, as well as the path followed by the visitor, are automatically saved by the Visite Plus system on a personal Web page accessible on the museum's Web site after the visit through the number of the bar-coded ticket or PDA. In this way, the visitor is able to analyze in more depth the subjects that were particularly interesting during the exhibition (through the provision of additional information) and to compare experiences with those of other visitors.

The fact that an important part of the content concerning the exhibition is accessible at home or in another context, after the actual visit, allows the visitor to focus more on experimentation and discovery while in the museum and to leave the more traditional didactic aspects for later. The Visite Plus system also offers the possibility of subscribing to a personalized periodical newsletter that focuses on a series of themes selected by the visitors at the moment of the registration from a list of available subjects. As well, a complete dossier of the exhibition can be provided. See Figure 9 for an example of the view of the exhibit from the personalized Web site. Each square corresponds to a content area in the exhibition. The squares that are in full color represent areas that have been accessed during the visit to the exhibition, while the white ones correspond to areas that have not been accessed.

Figure 9. Visite Plus exhibition Web site

In a similar manner, the London Science Museum's In touch project allows a record of a visitor's interaction with various interactive exhibits in the Wellcome Wing, including eye scan, voice, face and fingerprint recognition, photo editing, etc., to be recorded using fingerprint as an identifier, thus avoiding the need for any ticket. The results are made available as a Web site that can be accessed via the visitor's first name and birth date (http://www.sciencemuseumintouch.org.uk). Since the year 2000, when the project was originally implemented, Joe Cutting of the Science Museum reports that (as of January 2004) more than 400,000 Web pages have been created, of which around 8% have been accessed at least once. Figure 10 shows two screenshots from the exhibition itself, and Figure 11 shows example pages from the associated Web site.

Figure 10. London Science Museum Wellcome Wing "In touch" exhibition screenshots

Figure 11. London Science Museum Wellcome Wing "In touch" associated Web site

In order to simplify the system, reduce the operational problems that derive from such a large database, and increase the percentage of visitors using it, the museum has decided to replace the fingerprint method (which is not completely reliable in practice) by "an email it to me" option by the end of 2004. Every time a person wants to save one of the interactions, an email address will have to be provided. Thus there will be no more automatically generated personal pages for the visitors. However, the museum is thinking of including a link in the email to allow the visitor to set up a personal page if desired. In this way only those who are really interested will set up a page, and the museum will not have to maintain a huge and largely unused database.

The Getty Guide project is exploiting the same principle of visit and post-visit experience, and eventually also allowing pre-visit personalization. A new PDA is planned for launch in May 2004 to accompany the visitor in the museum by providing audio descriptions, images, some (even if limited) text and interactive maps. It allows the user to create a personal tour on the museum Web site by bookmarking (up to 50) objects, to save them by providing a username and a password, and then to download them on to the museum PDA during the actual visit. People can also bookmark objects of particular interests while in the museum and then inspect them on the museum kiosks where, by selecting a username and a password, they can be saved and then accessed through the Web site under My Getty - or emailed to themselves or somebody else.

Conclusion

In conclusion, there is clear benefit in taking the different needs of visitors into account when providing access to information and services in museums. However there are drawbacks because of difficulty and expense, and also problems in practical.use. For a more detailed overview, , see Filippini-Fantoni (2003a; 2003b). So far it seems that only a limited number of visitors take advantage of the benefits available through personalization, partly because the systems are not implemented in a clear and easy manner and partly because most visitors are either not ready for technology or not willing to invest time in it. Therefore it is important to remember that personalization should not be implemented for its own sake, but when and because it brings added value to the museum for a good percentage of museum visitors. Only then can the costs for investment and development be justified.

Some experts have warned against the use of personalization. For example, Nielson (1998) argues that personalization is over-rated, saying that good basic Web navigation is much more important. For example, it is helpful to consider different classes of use in the main home page, such as on-site visitors, the disabled, children, teachers, researchers, groups, etc., and to give each of these a relevant view of the resources that are available (Bowen & Bowen, 2000). Such usability issues are certainly important and relatively cheap to address with good design, but even Nielson admits that there are special cases were personalization is useful.

More recently, there have been further questions about the effectiveness of personalization (Festa, 2003; McGovern, 2003), despite the enthusiasm of some. For example, the costs may be up to four times those of a normal Web site; around a quarter of users may actually avoid personalized Web sites due to privacy concern; and only 8% are encouraged to revisit because of personalized facilities (Jupiter Research, 2003). This compares with 54% who consider fast-loading pages, and 52% for better navigation, as being important. However, other surveys indicate that personalization can be effective; for example in the field of downloadable music (Tam & Ho, 2003).

Thus it is recommended that museums use personalization on Web sites judiciously at the moment. There is a place for it in leading edge Web sites and for certain innovative facilities like advanced Web support for specific exhibits. It is an area that museums should certainly consider, but the costs should be weighed against the benefits. Of course, the costs are likely to decrease as commercial and open source support improve in this area. At the moment, not insignificant development effort is needed for such facilities, but in the future they could be increasingly packaged with standard Web support software as experience of what is useful and what is not useful is gained from practical experience.

Currently, quite a number of Web content management systems do include some personalization features. Ten entries in the Google directory entry for content management explicitly mention "personalization" (http://directory.google.com/Top/Computers/ Software/Internet/Site_Management/Content_Management/) as of January 2004. However, the features provided might not be what are required for museum applications since these systems are often designed with e-commerce in mind at the moment.

For leading national and international museums, personalization is certainly something to consider now if they have not already done so. For others, it may be best to wait till the costs decrease and the benefits to museums can be properly assessed in practice. Personalization may be best implemented at a meta-level currently. For example, a museum Web portal could assess the interests, expertise, background and physical location of a user to present a selection of museums, exhibitions, resources, news, etc., likely to appeal to that user. It would also be worth encouraging general portals like My Yahoo and My MSN to include more personalized content from museums.

For the future, research into areas like agent-based interfaces (Negroponte, 1995) and the semantic Web (Berners-Lee, 1999) could help to improve the effectiveness of personalization. To quote from Negroponte (1995) again: "In the post-information age, we [will] often have an audience the size of one. Everything is made to order, and the information is extremely personalized." Although museums and other organizations are making initial steps in this direction, we still have some way to go to reach Negroponte's Utopian digital dream.

Acknowledgements: We would like to thank the following people for providing information on the projects mentioned in this paper: Joe Cutting, Ann Borda and Alpay Beler of the Science Museum, London (In touch and Ingenious projects), Christina Olsen of the J. Paul Getty Museum (The Getty Guide), Roland Topalian and Joelle Le Marec of La Cité des Sciences et de l'Industrie (Visite Plus), Fabio Paternòò and Luisa Marucci of the Italian National Research Centre (the Carrara Marble Museum Web site). Thank you especially to Ann Borda, Joe Cutting and Roland Topalian for giving us permission to reproduce screenshots of their respective projects in this paper. We would also like to thank Kati Geber and the Canadian Heritage Information Network (CHIN) for providing the framework under which some of this research on personalization was undertaken.

References

Beardon, C. & S. Worden (1995). The virtual curator: multimedia technologies and the roles of museums. In E. Barrett & M. Redmond (Eds.), Contextual Media: multimedia and interpretation. Cambridge, MA: MIT Press, Series on Technical Communication, Multimedia, and Information Systems, pages 63-86. http://doi.acm.org/10.1145/212279.212291

Berners-Lee, T. (1999). Weaving the Web: The original design and ultimate destiny of the World Wide Web, by its inventor. HarperCollins. http://www.w3.org/People/Berners-Lee/Weaving/

Bonnet, M. (2002). Personalization of web services: opportunities and challenges. Ariadne, Issue 28, June. http://www.ariadne.ac.uk/issue28/personalization/

Bowen, J. P. (2002). Weaving the museum web: the Virtual Library museums pages. Program: Electronic Library and Information Systems, 36(4), 236-252.

Bowen, J. P. (2004). Cultural Heritage Online. Ability, Issue 53, pages 12-14, January. http://www.abilitymagazine.org.uk/features/2004/01/A53_Cover_story.pdf

Bowen, J. P. & J. S. M. Bowen (2000). The Web site of the UK Museum of the Year, 1999. In D. Bearman & J. Trant (Eds.), Proc. Museums and the Web 2000, Minneapolis, USA, 16-19 April 2000. Archives & Museum Informatics. http://www.archimuse.com/mw2000/papers/bowen/bowen.html

Bowen, J. P., M. Houghton & R. Bernier (2003). On-line museum discussion forums; What do we have? What do we need? In D. Bearman & J. Trant (Eds.), Proc. MW2003: Museums and the Web 2003, Charlotte, USA, 19-22 March. Archives & Museum Informatics. http://www.archimuse.com/mw2003/papers/bowen/bowen.html

Brusilovsky, P. & M. T. Maybury (2002). From adaptive hypermedia to the adaptive Web. Communications of the ACM, 45(5), 30-33, May. http://doi.acm.org/10.1145/506218.506239

Brusilovsky, P. & M. T. Maybury, Eds. (2002). Special Issue: The adaptive Web. Communications of the ACM, 45(5), 30-67, May. http://doi.acm.org/10.1145/506218.506238

Chevest, K., K. Mitchell & N. Davis (2002). The role of adaptive hypermedia in a context-aware tourist GUIDE. Communications of the ACM, 45(5), 47-51, May. http://doi.acm.org/10.1145/506218.506244

Eirinaki, M. & M. Vazirgiannis (2003). Web mining for Web personalization. ACM Transactions on Internet Technology, 3(1), 1-27, February. http://doi.acm.org/10.1145/643477.643478

Festa, P. (2003). Report Slams Web Personalization. CNET News.com, 14 October. http://news.com.com/2100-1038_3-5090716.html

Filippini-Fantoni, S. (2003a). Museums with a personal touch. In J. Hemsley, V. Cappellini & G. Stanke (Eds.), EVA 2003 London Conference Proceedings, University College London, UK, 22-26 July, pages s25:1-10.

Filippini-Fantoni, S. (2003b). Personalization through IT in museums. ICHIM 03: International Cultural Heritage Informatics Meeting, L'Ecole du Louvre, Paris, France, 8-12 September.

Filippini-Fantoni, S. & J. P. Bowen (2004). Online: Personalization and customization in museum Web sites and IT. New Heritage, 01.04, February. To appear.

Hitzeman, J., C. Mellish & J. Oberlander (1997). Dynamic generation of museum Web pages: the Intelligent Labelling Explorer. Archives and Museum Informatics, 11(2), 107-115. http://www.hcrc.ed.ac.uk/Site/HITZJ971.html

Jupiter Research (2003). Beyond the Personalization Myth: Cost Effective Alternatives to Influence Intent. Jupiter Media, USA, 14 October. http://www.internet.com/corporate/releases/03.10.14-newjupresearch.html

Kobsa, A., J. Koeneman & W. Pohl (2001). Personalized hypermedia presentation techniques for improving online customer relationship. The Knowledge Engineering Review, 16(2), 111-155.

Linden, G., B. Smith & J. York (2003). Amazon.com recommendations: item-to-item collaborative filtering. IEEE Internet Computing, 7(1), 76-88, January-February. http://csdl.computer.org/comp/mags/ic/2003/01/w1076abs.htm

Lynch, C. (2000). From automation to transformation: forty years of libraries and information technology in higher education", Educause Review, 35(1), 60-68, January/February. http://www.educause.edu/pub/er/erm00/pp060068.pdf

Manber, U., A. Patel & J. Robison (2000). Experience with personalization on Yahoo! Communications of the ACM, 43(8), 35-40, August. http://doi.acm.org/10.1145/345124.345136

McGovern, G. (2003). Why personalization hasn't worked. New Thinking, 20 October. http://www.gerrymcgovern.com/nt/2003/nt_2003_10_20_personalization.htm

Mobascher, B., R. Cooley & J. Srivastava (2000). Automatic personalization based on Web usage mining. Communications of the ACM, 43(8), 142-151, August. http://doi.acm.org/10.1145/345124.345169

Negroponte, N. (1995). Being Digital. Knopf.

Nielsen, J. (1998). Personalization is over-rated. Alertbox, 4 October 4. http://www.useit.com/alertbox/981004.html

Oberlander, J., C. Mellish, M. O'Donnell & A. Knott (1997). Exploring a Gallery with Intelligent Labels. Proc. ICHIM'97: 4th International Conference on Hypermedia and Interactivity in Museums, pages 153-161. University of Edinburgh, Scotland, UK. http://citeseer.nj.nec.com/oberlander97exploring.html

Paternò, F. & C. Mancini (1999). Designing Web user interfaces adaptable to different types of use. In D. Bearman & J. Trant (Eds.), 1999 Museums and the Web. Selected Paper from an International Conference. Archives and Museum Informatics. http://www.archimuse.com/mw99/papers/paterno/paterno.html

Peppers, D. & M. Rogers (1996). The One-to-One Future: Building Relationships One Customer at a Time. Judy Piatkus (Publishers) Ltd.

Picard, R W. (1997). Affective Computing. Cambridge, MA: MIT Press.

Pine II, B. J. (1999). Mass Customization: The New Frontier in Business Competition. Cambridge, MA: Harvard Business School Press.

Riecken, D., Ed. (2000). Personalized views of personalization. Communications of the ACM, 43(8), 26-158, August. http://doi.acm.org/10.1145/345124.345133

Russell, J. (2003). Making it personal: information that adapts to the reader. Proc. 21st Annual International Conference on Documentation, San Francisco, USA, pages 160-166, October. ACM Press. http://doi.acm.org/10.1145/643477.643478

Tam, K. Y. & S. Y. Ho (2003). Web personalization: is it effective? IT Professional, 5(5), 53-57, September-October. http://csdl.computer.org/comp/mags/it/2003/05/f5053abs.htm